Detect-phishing-websites-using-ML

This project is a simple example which trains the model to predict phishing websites. Phishing websites are fake websites which try to gain the trust of users to steal private data of users.

• Best accuracy score - 97.0% using Random forest method
• Worst accuract score - 48.5% using One class svm method

Requirements

• Scikit-learn (sklearn)
• Numpy

Requirements can be installed by executing pip install -r requirements.txt

Data set

The data set for training has been taken from UCI archive

Execution

• python classifier.py to check the accuracy of the script.
• python classifier.py google.com to check whether google.com is phishing website or not.

Parameters in dataset

Each value in the dataset contains all these elements and all are seperated by a comma.

1. having_IP_Address { -1,1 }
2. URL_Length { 1,0,-1 }
3. Shortining_Service { 1,-1 }
4. having_At_Symbol { 1,-1 }
5. double_slash_redirecting { -1,1 }
6. Prefix_Suffix { -1,1 }
7. having_Sub_Domain { -1,0,1 }
8. SSLfinal_State { -1,1,0 }
9. Domain_registeration_length { -1,1 }
10. Favicon { 1,-1 }
11. port { 1,-1 }
12. HTTPS_token { -1,1 }
13. Request_URL { 1,-1 }
14. URL_of_Anchor { -1,0,1 }
15. Links_in_tags { 1,-1,0 }
16. SFH { -1,1,0 }
17. Submitting_to_email { -1,1 }
18. Abnormal_URL { -1,1 }
19. Redirect { 0,1 }
20. on_mouseover { 1,-1 }
21. RightClick { 1,-1 }
22. popUpWidnow { 1,-1 }
23. Iframe { 1,-1 }
24. age_of_domain { -1,1 }
25. DNSRecord { -1,1 }
26. web_traffic { -1,0,1 }
27. Page_Rank { -1,1 }
28. Google_Index { 1,-1 }
29. Links_pointing_to_page { 1,0,-1 }
30. Statistical_report { -1,1 }
31. Result { -1,1 }